Digital Dashbord

parameters

voltage,

display for

4.5" screen is too small to be practical for a car dash even if the information parameters are sequenced or prioritised in some way, especially if used for radio and/or GPS simultaneously with speed.

Good examples of digital and part digital displays can be found in agricultural tractors. An example of full digital display of multiple parameters is the Ford Series Ten generation II range fitted with SuperQ cab. This kind of display has long gone out of production due to the difficulty of reading digital rev counters and speedometers, let alone the bar type fuel and temperature gauges. A more successful dash of that era is that fitted to JD55 series machines which have LCD displays which incorporate large rotary rev counters in LC. Modern instruments tend to combine precision digitally signalled but 'clock-face' major instruments with one or more TFT or LCD displays which provide less needed information only when indicating outside normal operating parameters. It is also ergonomically sensible to have separate displays for indicating operating mode parameters and operating condition parameters. Have a look at this link for the state of the art. Please explore the whole site carefully to ensure a full overview of both dash and side panel displays. The side panel incorporates a full colour display of 'mode' parameters while defaulting to operational 'condition' as needed. The main dash panel mainly displays 'conditions' such as speed and revs and fuel remaining.

Huw

"Huw" wrote

Woops! Forgot to include the link

Huw

Dear Friends,

I am honoured by the reply that I have got to my quary. Well the need is to design and should be able to retrofit to ANY car. yes the input transduce may have to be designed saperately for each car depending on the kind of input i get. however India had very little choice till now and designing for those 2 or 3 types of car is ok. All new generation cars that are now available in India has some kind of a Microcontroller which might have some port or may be the parameters can be tapped from it. I will go through every car's controller may be a bit later. Right now designing it for a car which has all analogue input to the dashboard and tap it use a DAC and display it.

looking forward to more inputs.

shanks

but where is the link address.

shanks

I say again : So ?

If you're concerned about breaking the speed limit for two whole seconds, you're hardly likely to be "having a bit of fun, accelerating quite hard" :)

As a matter of fact, the receiver doesn't matter so much.

It doesn't measure speed, it measures absolute position and time. Now, you can use the change in position with time to calculate speed and this is indeed what a receiver does, but that's got some untoward effects due to the maths involved.

The basics are that your receiver gives you your position, the time (and also a derived speed) once per second and includes an error in position which is usually of the order of "give or take several meters."

For the interval of one second, your position might be five meters west at the beginning, and five meters east at the end. This is ten meters difference in that one second, and that's typical rather than sometimes.

So if you were doing 90mph (or 40 m/s) you would have a speedo that usually varied once each second to a figure that might be 80 mph to 100 mph using good modern receivers, or 45-135 mph on the older generation (or with waas turned off or not available) and maybe 20 to 160 mph for the cheap kit. So it goes "20", and stays there for a bit, then "110", and wait a little, it's "45", and so on.

Clearly, this is no use, so you would use some sort of algorithm to smooth the readings and produce an average in the middle of them.

Because your GPS receiver is only able to assess position once per second, due to the way the satellites only transmit pulses that often, however you smooth will depend on taking a number of seconds of previous information and averaging them out in some fashion - which means the reading at any given moment will be a number of seconds out of date when the actual speed is changing.

Now, this works fine with constant speed but the acceleration situation is very different. Let's imagine a simple algorithm which takes the change in position over the last five seconds, and divides that by five to get the current speed. This isn't a bad algorithm, and you won't get higher *accuracy* by improving on it, you will only vary the display figure to either be more averaged or less averaged.

Once you are into this averaging situation, changes in speed are necessarily reduced for the display since you are including figures that are now definitely wrong.

I'll illustrate this with figures. Let's have a quick car accelerating at half a "g" and ignoring wind resistance. There are four columns, time in seconds, position from the start point in meters, actual speed at the end of that second, and derived speed based on the difference in position at the end of that second and the position five seconds ago, divided by five seconds, both in meters per second. The fifth column is acceleration in meters per seconds squared.

The equations for excel are 1,0,0,0,5 for the first row (starting with row 11 as you need to refer back (a11 - e11), then for the second row (row 12), =a11+1, =0.5*e12*a12*a12, =e12*a12, = (b12-b7)/5, =e11

Time Dist Speed Reads Acc

0 0 0 0 5 1 2.5 5 0.5 5 2 10 10 2 5 3 22.5 15 4.5 5 4 40 20 8 5 5 62.5 25 12.5 5 6 90 30 17.5 5 7 122.5 35 22.5 5 8 160 40 27.5 5 9 202.5 45 32.5 5 10 250 50 37.5 5 10.78 290.521 53.9 40.1042 5

The last row is one I adjusted the time manually, so that the GPS would be reading 90 mph at that time, which turns out to be after 10.78 seconds (it actually reached 90 mph in 8 seconds).

Basically, if you accelerated from rest at half a g and the algorithm was a five second average, and there were no errors in measurement in the GPS then you would get a reading of 90 mph, once you were actually travelling at 120 mph.

Once you cease accelerating, the reading rises to be equal to the actual speed in about five seconds. Add in the error figures and you'll never really know your speed until things have settled down, a few seconds on, which is probably ok for cruise control but not much use for speed camera control or accelerating to the speed limit and then levelling off by a human driver using the speedo to judge when to stop accelerating.

What it boils down to is this:

The refresh rate of a GPS speedo is at best, once per second. The raw accuracy is +/- 10 meters overall on the best kit, which translates to +/- 25 mph, and the less good kit gives you +/- 30 meters, or +/- 70 mph for that second. To smooth things out, the refresh rate means the speedo must lag behind the actual changes in speed by a number of seconds, and will therefore be inaccurate when speed is actually changing. Quality of the receiver can't change these basic facts, although an updated form of GPS could one day be launched with greater precision on position, or frequency of update, and this would make it possible to improve things in this respect.

Now, to address the original question, an electronic dashboard. The obvious way to get speed is to measure it directly, e.g. to have a magnet welded to the prop shaft next to a coil and count induced pulses which relate to how many times the propshaft / axle / wheel has turned during that period of time.

A high tech, fancy, and non-tyre dependent method which can be incorporated into the dashboard, is to use a two axis accelerometer. This updates far more frequently and tells you your acceleration at any point in time. You can get single axis, two or three axis, single will do for a simple solution but two means the speedo continues to work when you are going sideways and also allows for more sophisticated applications like tracking the motion against a map as you can tell when the car changes direction as well as speed. Three axis is probably overkill, but as the price is much the same it may be worth using this type.

Accelerometers give rapid and accurate information about changes in speed, and are extremely useful for car measurements because you are directly measuring acceleration and deceleration (engine power output, braking effectiveness, plus tyres in both cases.)

They're not perfect for speed, though, a small error in measuring acceleration makes a large error in speed, although you are talking very accurate devices there so it's not a problem, also the error will hang around in the system, e.g. if you accelerated to 40 m/s, and only decelerated by 39.5 m/s, you're now travelling at 0.5 m/s even when stationary. Do the same thing, a few times, and you're soon doing "45" m/s. As constant speed is 0 acceleration the device does not know. Now combine the inertial system (accelerometers) and the GPS system (absolute position) and you have the best of both worlds, instant and precise measures of acceleration, plus a long term stable position reference to calibrate your sensors and correct errors in the system from. Add in a computer (PPC is a good, cheap type and has decent storage options like CF cards which survive fine in a dashboard environment) and maps and you have a dashboard navigation system that also works in tunnels / inside a ship. Combine it with GPRS telephony and you can track your car's position and have a program send you email / phone message when the car moves off from the last position, so not an easy car to steal, for instance.

It is a problem - the error in velocity calculated by integration of acceleration increases linearly with time, so a device calibrated for -2g to +2g, with a dynamic range of 5000, even if properly calibrated will result in a velocity error of 28.24 m/s (63.2 mph)after 1 hour.

It is possible to measure this error and compensate for it, but there is a more fundamental problem, which is that it is impossible to differentiate between acceleration due to the car changing speed and acceleration due to e.g. climbing a hill. If you drove up a 10% hill then your indicated speed would be in error by approx. 2mph /second.

Agreed. This combination is used in many self-guiding missiles.

I looked at this problem some time ago, and ultimately decided to use a doppler ultrasound measurement of road speed, which doesn't suffer from accumulative integration errors and can be calibrated to any arbitrary degree of accuracy required.

This is certainly worth using a 3 axis version for, if that's likely to be coming into the equations. And I agree that the precision will add errors which build up with significant time and need to be compensated out somewhere.

That's an interesting idea, tell me more :)

I'm guessing you point a sonar / radar gun at the road and measure the returns?

Sort of. I used 40KHz Ultransonic transducers, since they are cheap. Pointed at ground under car, at known angle to road(theta), mix RX signal with TX signal - gives (Ftx+Frx) and (Ftx-Frx) signals, filter off (Ftx+Frx) componenent (~80KHz), leaves difference freq (Ftx-Frx)= doppler shift. You can measure this any number of ways. doppler shift=Ftx*(Vcar/Vsound), some simple rearranging gives Vcar, although you need to account for cos(theta) term for angle of incidence.

It gets a bit more complex to get it really accurate, but that's the basic principle, anyway. Sorry if this is incomprehensible, but I'm in a rush...